Has anyone come across proper documentation stating exactly why quadro cards are a prefered over fermi gtx for cg applications in general?
I haven't read anything like an FAQ on the matter, but I have been able to piece together the following bits from reading articles & posts over the years:
- GEForce cards have their geometry rates capped, while Quadro cards do not. Games are often more shader-heavy than geometry-heavy, while digital content creation (DCC) is usually the opposite. This allows the game-oriented GEForces to be clocked higher than the Quadros, boosting their pixel shading output, while still remaining within thermal and power limits. In other words, the GEForces are firmware-optimized for games.
- Quadros have more memory than their equivalent GEForce counterparts, usually on the order of 2-4x, simply because games impose rigid requirements on the textures and geometry they draw, whereas DCC packages cannot.
- Drivers for Quadros are optimized differently than GEForces. DCC mostly draw to multiple windowed viewports at a time, while games often draw to a single, fullscreen window. Quadro drivers optimize workstation-app GL features such as smooth lines, and tend to be more stable over time for workstation apps like Houdini.
- Marketing often dictates a certain feature set for workstation vs. gaming cards. Just like Houdini Escape is less expensive than Master but with fewer features, so the GEForce is cheaper than the Quadro. Examples of features that Quadros have that GEForces do not - ECC memory, dual DMA copy engines, quad-buffer stereo, half-rate FP64 math (vs. 1/4, 1/12), and faster geometry processing. Many of these are crucial to Quadro users.
Most workstation applications will still run on a GEForce card, just not as well. Here are some benchmarks:
hothardware [
hothardware.com],
Tom's Hardware [
tomshardware.com]. I suppose it's up to you as a consumer to judge if the performance and feature benefits are worth the extra expense.
